The presence in carcinomas of large numbers of myofibroblasts (MFs) originally suggested the concept of cancer as a non-healing wound, and experiments have now implicated these cells as contributing to cancer growth, invasion and metastasis. Recently our labs (Tycko collaborating with Wang) showed that cancerassociated myofibroblasts (CAFs) in human gastric carcinomas (GCAs) and in a mouse model of GCA have globally reduced DNA methylation and focal gains of promoter methylation, compared to normal MFs in the stomach. We now have substantial unpublished data indicating that these findings of altered DNA methylation in CAFs extend to pancreatic carcinoma (PnCA), and that the demethylating drug decitabine (5aza-dC) has strong anti-tumor activity in a mouse model of PnCA, mediated in part through its effects on the tumor-supporting activity of CAFs and in part via its direct effects on the malignant epithelial cells. To build on these findings we have 4 current objectives - all centered on DNA methylation as a therapeutic target in both the supportive CAFs and malignant epithelial cells of stroma-rich gastrointestinal cancers. First, we will carry out several types of molecular assays to ask whether CAFs accumulate global and genespecific alterations in DNA methylation in pancreatic and hepatocellular carcinomas. Second, we will use mixing/allografting experiments to ask whether the demethylating drug decitabine inhibits the tumorsupporting function of CAFs, and we will profile gene expression and characterize mesenchymal stem cell markers in response to this drug to gain insights to its mechanisms of action against CAFs. Third, we will use in vivo mouse models of PnCA to test whether decitabine, alone and in combination with other agents, can prevent tumor progression and lead to regression of established PnCA tumors. This objective is linked to a Phase 1 clinical trial of decitabine in humans with PnCA, for which we will carry out the laboratory-based pharmacodynamic studies. Our fourth and last objective is to use a genetic strategy in mice to test whether deletion of the maintenance methyltransferase gene Dnmtl, specifically in MFs, can slow or prevent liver fibrosis and cancer in a chemical carcinogenesis model of hepatocellular carcinoma.